CN104661021A - Quality assessment method and device for video streaming - Google Patents

Abstract

The invention provides a video stream quality evaluation method and device. The method for evaluating the quality of the video stream includes: obtaining a first compressed video stream generated by processing an original video sequence; the first compressed video stream carries a sequence identifier and an image sequence number corresponding to the original video sequence; according to the The sequence identifier is used to obtain the original video sequence corresponding to the first compressed video stream; according to the sequence identifier and the image sequence number, the original compressed video stream corresponding to the original video sequence is obtained; according to the first compressed video and the original compressed video stream, and evaluate the video quality of the first compressed video stream. The present invention can realize full-reference evaluation of video objective quality at different processing stages of video streams without increasing network transmission burden.

Description

Method and device for evaluating video stream quality

technical field

The invention relates to multimedia communication technology in the field of communication, in particular to a method and device for evaluating the quality of video streams.

Background technique

At present, with the rapid development of the Internet and mobile communication networks, people's demand for video services continues to increase, and services such as video surveillance, video conferencing, and online video playback are growing day by day. The basic feature shared by all kinds of video processing and transmission systems is: the video stream generated by one node is transmitted to another node through the network. The links that affect the quality of the video stream include the generation, transmission and reconstruction stages of the video stream. Due to the huge amount of original video source data, lossy compression standards are usually used in the generation stage of video streams to greatly reduce the amount of data to be transmitted, and at the same time bring about a decline in video quality; network packet loss, delay, and jitter during network transmission Such problems will also bring about the decline of video quality; in addition, in the video stream reconstruction stage, factors such as the display quality of the terminal device and the lighting environment will also affect the video quality. Since the above-mentioned various types of video degradation factors are different, it brings great difficulties to the video quality evaluation of the video receiving end.

The current evaluation methods for video quality can be divided into Video Subjective Quality Assessment (VSQA for short) and Video Objective Quality Assessment (Video Objective Quality Assessment for short) according to whether the evaluation conclusion is given by human eye observation. VOQA).

Video subjective quality assessment is to play a series of test video sequences in the test environment specified by the tester according to international standards (such as ITU-R BT 500), and let the testees give subjective ratings on the quality of these test video sequences. Since the subjective rating given by the test video sequence is the perception value of the test video under human vision, the result of the subjective evaluation is considered to be accurate. However, the process of subjective evaluation is cumbersome and time-consuming, and the test results obtained from the evaluation are not scalable, so it cannot be used in fields with high real-time requirements.

The objective video quality assessment builds a suitable mathematical model by analyzing the original video sequence and the test video sequence, and then extracts the key data parameters that characterize the video damage, and uses these parameters as the evaluation value of the objective video quality. This method is simple and fast, and can meet the Real-time requirements have been widely used.

Video objective quality assessment methods are further divided into full-reference video quality assessment methods, partial-reference video quality assessment methods and no-reference video quality assessment methods. The video quality assessment methods of full-reference type and partial-reference type usually need to refer to all or part of the information of the original video sequence, but in practical applications, it is often difficult for the receiver to obtain the information of the original video sequence. The non-reference type video quality assessment method does not need to transmit any information of the original video sequence, and can directly estimate the degree of video distortion based on some distortion characteristics of the video stream received by the receiving end. This type of method is still in the research stage , and cannot accurately obtain the degree of real video distortion, and its application has certain limitations.

In the practical application of various network video services, what is needed is an evaluation method that is easy to configure, simple and efficient, and can accurately detect the objective quality of video in the stages of video stream generation, transmission, and reconstruction.

Contents of the invention

The technical problem to be solved by the present invention is to provide a video stream quality evaluation method and device, which can realize full reference evaluation of video objective quality in different processing stages of the video stream without increasing the burden of network transmission.

The method for evaluating the quality of the video stream includes:

Obtaining a first compressed video stream generated by processing the original video sequence; the first compressed video stream carries a sequence identifier and an image sequence number corresponding to the original video sequence;

Acquiring the original video sequence corresponding to the first compressed video stream according to the sequence identifier;

Acquiring an original compressed video stream corresponding to the original video sequence according to the sequence identifier and the image sequence number;

Evaluate the video quality of the first compressed video stream according to the first compressed video stream and the original compressed video stream.

The present invention also provides a method for evaluating the quality of video streams, including:

Obtaining an original video sequence, a sequence identifier corresponding to the original video sequence, and an image sequence number corresponding to the original video sequence;

An original compressed video stream corresponding to the original video sequence is generated according to the sequence identifier and the image sequence number.

The present invention also provides a quality assessment device for video streams, including:

The first acquisition unit acquires a first compressed video stream generated by processing an original video sequence; the first compressed video stream carries a sequence identifier and an image sequence number corresponding to the original video sequence;

A second obtaining unit, according to the sequence identifier, obtains the original video sequence corresponding to the first compressed video stream;

A third obtaining unit, according to the sequence identifier and the image sequence number, obtains the original compressed video stream corresponding to the original video sequence;

The first evaluation unit evaluates the video quality of the first compressed video stream according to the first compressed video stream and the original compressed video stream.

The present invention also provides a quality assessment device for video streams, including:

An acquisition unit, configured to acquire an original video sequence, a sequence identifier corresponding to the original video sequence, and an image sequence number corresponding to the original video sequence;

The generating unit generates an original compressed video stream corresponding to the original video sequence according to the sequence identifier and the image sequence number.

The beneficial effects of above-mentioned technical scheme of the present invention are as follows:

The present invention can identify the corresponding compressed video code stream and original video sequence at the receiving end without increasing the burden of network transmission, so as to realize the full reference evaluation of the video objective quality in different processing stages of the video stream respectively, and has flexible application , high precision, objective evaluation and other characteristics, can be widely used in the field of video.

Description of drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the present invention will be further described below in conjunction with the accompanying drawings and implementation examples. Obviously, the embodiments described in the present invention are part of the embodiments of the present invention. Based on the embodiments described in the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative efforts belong to The protection scope of the present invention.

FIG. 1 is a schematic flowchart of a video stream quality assessment method provided by an embodiment of the present invention;

FIG. 2 is a schematic connection diagram of a video stream quality assessment device provided by an embodiment of the present invention;

FIG. 3 is a schematic diagram of an apparatus for evaluating the objective quality of a video code stream according to an embodiment of the present invention;

FIG. 4 is a schematic diagram of a method of a video stream generation unit provided by an embodiment of the present invention;

5 is a schematic diagram of a method of a video acquisition and frame information identification unit provided by an embodiment of the present invention;

FIG. 6 is a schematic diagram of a method of a video stream objective quality calculation unit provided by an embodiment of the present invention;

FIG. 7 is a schematic diagram of an application scenario for video stream transmission quality assessment provided by an embodiment of the present invention;

FIG. 8 is a schematic diagram of an application scenario for evaluating compression quality of a video encoding device according to an embodiment of the present invention.

FIG. 9 is a schematic diagram of an application scenario for video stream reconstruction quality assessment provided by an embodiment of the present invention.

Detailed ways

In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, the following will describe in detail with reference to the drawings and specific embodiments.

As shown in Figure 1, a kind of quality assessment method of video stream is provided for the present invention, comprises:

Step 11: Obtain a first compressed video stream generated by processing the original video sequence; the first compressed video stream carries a sequence identifier and an image sequence number corresponding to the original video sequence. That is to say, each video sequence is composed of a series of video images, and playing at a certain frame rate will form a moving image. The present invention has multiple original video sequences, and different original video sequences correspond to different sequence identifiers. The same original video sequence corresponds to different video images, and different video images of the same original video sequence correspond to different image sequence numbers.

Wherein, the processing of the original video sequence includes: encoding processing of the original video sequence; or, transmission processing of the original compressed video stream generated by the original video sequence; The decoding process of the raw compressed video stream generated by the video sequence. Correspondingly, step 11 is: Step 11A, receiving the video of the original compressed video stream after transmission processing as the first compressed video stream; or, step 11B, receiving the decoded video of the original compressed video stream as the first compressed video stream A compressed video stream; or, step 11C, collect the video displayed on the original compressed video stream as the first compressed video stream. Taking the processing as transmission processing as an example, the original video sequence and the corresponding original compressed video stream are located at the sending end, and the first video sequence and the corresponding first compressed video stream are located at the receiving end.

Step 12, according to the sequence identification, obtain the original video sequence corresponding to the first compressed video stream;

Step 13, according to the sequence identifier and image sequence number, obtain the original compressed video stream corresponding to the original video sequence;

Step 14: Evaluate the video quality of the first compressed video stream according to the first compressed video stream and the original compressed video stream.

The method also includes:

Step 15, obtaining the first video sequence corresponding to the first compressed video stream; Step 15 is specifically:

Step 15A, decoding the first compressed video stream to generate a first video sequence; or

Step 15B, collecting video displayed on the original compressed video stream to generate a first video sequence.

Step 16: Evaluate the video quality of the first video sequence according to the first video sequence and the original video sequence.

Optionally, before step 16, the method also includes:

Step 16A, judging whether there is frame loss in the first video sequence:

Step 16B, if there is a frame loss, make up the lost video frame, and generate the first video sequence after the frame complement;

Step 16 is specifically: according to the first video sequence after frame complementation and the original video sequence, evaluate the video quality of the first video sequence after frame complementation.

In one embodiment, step 12 includes:

Step 121A, extracting the sequence identifier carried by the first compressed video stream;

Step 122A, according to the corresponding relationship between the sequence identifier and the original video sequence, obtain the original video sequence corresponding to the sequence identifier.

In another embodiment, step 12 includes:

Step 121B, extracting the sequence identifier carried by the first compressed video stream;

Step 122B, according to the sequence identifier, generate the original video sequence corresponding to the sequence identifier.

In one embodiment, step 13 includes:

Step 131A, extracting the sequence identifier and image sequence number carried by the first compressed video stream;

Step 132A: Acquire the original compressed video stream corresponding to the sequence ID and the image serial number according to the correspondence between the sequence ID, the image serial number and the original compressed video stream.

In another embodiment, step 13 includes:

Step 131B, extracting the sequence identifier and image sequence number carried by the first compressed video stream;

Step 131B, superimposing the sequence identifier and the image sequence number on the frames of the original video sequence to generate an original compressed video stream.

Step 14 is specifically: Step 141, according to the first compressed video stream and the original compressed video stream, calculate the video packet loss rate of the first compressed video stream relative to the original compressed video stream.

Or, step 14 is specifically: step 142, according to the first compressed video stream and the original compressed video stream, calculate the video source bit error rate of the first compressed video stream relative to the original compressed video stream .

Step 141 is specifically:

Step 142 is specifically:

The calculation method of the number of erroneous bits is: the number of erroneous bits can be determined after the first compressed video stream and the original compressed video stream are directly compared with each other.

Step 16 is specifically: Step 161, according to the first video sequence and the original video sequence, calculate the mean square error of the first video sequence relative to the original video sequence;

Alternatively, step 16 specifically includes: step 162, calculating the peak signal-to-noise ratio of the first video sequence relative to the original video sequence according to the first video sequence and the original video sequence;

Alternatively, step 16 specifically includes: step 163, calculating, according to the first video sequence and the original video sequence, an average structural similarity value of the first video sequence relative to the original video sequence.

Step 161 is specifically:

$\mathrm{<span\; class="notranslate">\; MSE</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; N\; M</span>}}\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; Y</span>}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; ]</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>$

in, square error; the frame resolution of the first compressed video stream is M pixel × N pixel, and X (i, j) is the pixel value at point (i, j) of one frame image of the original video sequence; Y(i, j) represents the pixel value at point (i, j) of the corresponding frame image of the one frame image in the first video sequence; the pixel value may be grayscale or color difference.

Step 162 is specifically:

$\mathrm{<span\; class="notranslate">\; PSNR</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 10</span>}\mathrm{<span\; class="notranslate">\; log</span>}<span\; class="notranslate">\; [</span>\frac{\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; E.</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; Y</span>}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; ]</span>}<span\; class="notranslate">\; ]</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 4</span>}<span\; class="notranslate">\; )</span>$

Wherein, PSNR is the peak signal-to-noise ratio, the frame resolution of the first compressed video stream is M pixels×N pixels, and X(i, j) is one of the frames of the original video sequence at point (i, j ) at the pixel value; Y(i, j) represents the pixel value at point (i, j) of the corresponding frame image of the one frame image in the first video sequence; E is the first compressed video The peak amplitude of the stream at the sampling condition of the reserved bits;

Step 163 is specifically:

$\mathrm{<span\; class="notranslate">\; MS\; SIM</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; m</span>}}\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{\mathrm{<span\; class="notranslate">\; m</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; SSIM</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 5</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>$

Wherein, SSIM is a structural similarity parameter, and k is the sequence number of a local window of one of the frame images of the first video sequence; M is the total number of local windows of one of the frame images of the first video sequence; xk and yk are the content of the video frame of the k-th local window; the content is the collective name of all digital images in the window.

SSIM is calculated according to the following formula;

SSIM＝[l(x,y)] ^α ·[c(x,y)] ^β ·[s(x,y)] ^γ (6);

Among them, α, β, γ>0, are the weighting coefficients of brightness comparison function l(x,y), contrast comparison function c(x,y) and structure information comparison function s(x,y) respectively;

$\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 2</span>}{\mathrm{<span\; class="notranslate">\; \&mu;</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}{\mathrm{<span\; class="notranslate">\; \&mu;</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}}{{{\mathrm{<span\; class="notranslate">\; \&mu;</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{{\mathrm{<span\; class="notranslate">\; \&mu;</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; K</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>$

$\mathrm{<span\; class="notranslate">\; c</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 2</span>}{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}}{{{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; K</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 8</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>$

$\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; xy</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}}{{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; xy</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 9</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>$

in, Indicates the average brightness of the original video sequence; N is the total number of frames of the original video sequence.

Indicates the average brightness of the first video sequence;

Indicates the standard deviation of the original video sequence,

represents the standard deviation of the first video sequence;

Represent the covariance of the original video sequence and the first video sequence;

C ₁ , C ₂ , and C ₃ are constants;

K ₁ , K ₂ <<1, L represents the dynamic variation range of the pixel value.

As shown in Figure 2, the present invention also provides a video stream quality assessment device, including:

The first acquisition unit 21 is configured to acquire a first compressed video stream generated by processing the original video sequence; the first compressed video stream carries a sequence identifier and an image sequence number corresponding to the original video sequence;

The second obtaining unit 22 obtains the original video sequence corresponding to the first compressed video stream according to the sequence identifier;

The third acquisition unit 23 is to acquire the original compressed video stream corresponding to the original video sequence according to the sequence identifier and the image sequence number;

The first evaluation unit 24 evaluates the video quality of the first compressed video stream according to the first compressed video stream and the original compressed video stream.

The device also includes:

The fourth acquiring unit 25 is configured to acquire a first video sequence corresponding to the first compressed video stream;

The second evaluation unit 26 evaluates the video quality of the first video sequence according to the first video sequence and the original video sequence.

The first acquisition unit 21 includes:

The transmission module receives the video of the original compressed video stream after transmission processing as the first compressed video stream;

The decoding module receives the decoded video of the original compressed video stream as the first compressed video stream; or

The first collection module collects the video displayed on the original compressed video stream as the first compressed video stream.

The fourth acquisition unit 25 includes:

A decoding module, configured to decode the first compressed video stream to generate a first video sequence; or

The second collection module collects the video displayed on the original compressed video stream to generate a first video sequence.

The second acquisition unit 22 includes:

The first extraction module extracts the sequence identifier carried by the first compressed video stream;

The first acquiring module acquires the original video sequence corresponding to the sequence identifier according to the correspondence between the sequence identifier and the original video sequence.

Optionally, the second acquisition unit 22 includes:

The second extraction module extracts the sequence identifier carried by the first compressed video stream;

The first generating module generates an original video sequence corresponding to the sequence identifier according to the sequence identifier.

The third acquisition unit 23 includes:

The third extraction module extracts the sequence identifier and image sequence number carried by the first compressed video stream;

The second acquiring module acquires the original compressed video stream corresponding to the sequence ID and the image serial number according to the corresponding relationship between the sequence ID and the image serial number and the original compressed video stream.

Optionally, the third acquisition unit 23 includes:

The fourth extraction module extracts the sequence identifier and image sequence number carried by the first compressed video stream;

The second generation module superimposes the sequence identifier and the image sequence number on the frame of the original video sequence to generate an original compressed video stream.

Through different configuration methods, the embodiment of the present invention can identify the corresponding compressed video code stream and original video sequence at the receiving end without increasing the network transmission burden, so as to realize the generation, transmission and reconstruction phases of the video stream respectively. The full-reference evaluation of objective video quality has the characteristics of flexible application, high precision, and objective evaluation, and can be widely used in the video field.

The application scenarios of the method of the present invention are described below.

As shown in Figure 4, the present invention provides a kind of method for generating a test-specific video stream (equal to the above-mentioned original compressed video stream), including:

First, the original video test sequence (equivalent to the above-mentioned original video sequence) is divided into four types according to the lens motion characteristics: still, fast moving, slow moving and zooming, and all sequences are numbered alphabetically (equal to the above sequence identification ), the number of test sequences and encoding control parameters can be determined according to the specific test environment.

Then, according to the selected video compression method and reference frame structure, the picture sequence number (Picture Order Count, POC for short) of the single test sequence is numbered (equivalent to the above-mentioned picture sequence number), and in each original video The sequence letter number and POC number are superimposed on the specific position of the frame (the position can be determined according to the actual situation).

Select the required video compression standard, and generate a test-specific compressed video stream according to the encoding parameter table recommended by the standard (mainly including encoding control parameters such as grade level, quantization step size, and preset bit rate), and record the generated The average peak signal-to-noise ratio and bit rate of the video stream.

The embodiment of the present invention also provides a full-reference evaluation device for the objective quality of video streams, which can be used for testing two client terminals that need to communicate with each other. It includes: a video stream generating unit, a video stream receiving and analyzing unit, a video stream reconstructing and displaying unit, a video collecting and frame information identifying unit, and a video objective quality calculating unit. Among them, the video stream generation unit is connected to the video stream reconstruction and display unit, and the video objective quality calculation unit; the video stream reception and analysis unit is connected to the video stream reconstruction and display unit; the video stream reconstruction and display unit is connected to the video acquisition and frame information identification unit; The acquisition and frame information identification unit is connected to the video objective quality calculation unit; the video objective quality calculation unit is connected to the video stream generation unit.

The video stream generation unit deployed at the sending end and the receiving end at the same time generates a test-specific compressed video stream;

The video stream receiving and analyzing unit, after receiving the video stream at the receiving end, extracts the encoding grade level, quantization step size, and preset bit rate in the Network Abstract Layer Unit (NALU) in the video stream Parameter control parameters and POC and other parameters are analyzed, and at the same time, according to the received parameters, a compressed video stream consistent with the currently received video stream can be generated synchronously for use by the video stream objective quality calculation unit;

The video stream reconstruction and display unit completes the video stream decoding and display functions;

The video acquisition and frame information identification unit collects the locally acquired video sequence through the acquisition of the video display unit and the image signal identification of the designated area, and identifies the currently received video sequence number and POC number for use by the video objective quality calculation unit;

The video objective quality calculation unit, through the detection and identification of the currently received video stream, uses the video stream generation unit at the receiving end to synchronously generate a consistent video stream for the evaluation and calculation of the video objective quality; or directly extract the data stored at the receiving end Consistent video streams and original video data are used for objective video quality evaluation calculations; at the same time, consistent video code streams can also be used to calculate objective quality indicators such as packet loss rate and bit error rate of video code streams during channel transmission.

The beneficial effects produced by the present invention are:

On the one hand, the present invention first directly superimposes the type information of the video sequence and the corresponding image sequence number as a part of the test video sequence image, which facilitates the identification of video information at the receiving end; at the same time, it does not need to transmit the original video stream separately and will not In the case of increasing the burden of network transmission, the objective quality detection of video streams under full reference conditions can be realized.

On the other hand, the device disclosed in the present invention has the functions of original video stream output, compressed video stream output and received video stream input processing functions at the same time, so a single video communication node can be tested independently; The test environment is configured for the generation, transmission and reconstruction stages of video streams, with a more flexible configuration method.

On the other hand, the present invention can objectively and quantitatively analyze the video quality degradation introduced in the process of video compression, transmission and reconstruction, avoiding the subjectivity introduced by subjective testing methods; at the same time, authoritative objective evaluation indicators can be used to accurately describe the video stream quality and the reliability of the video transmission system. The invention is not only applicable to various video communication systems, but also can be used for equipment evaluation of video collection and encoding systems.

Examples of the present invention are described below.

Embodiment one:

This embodiment is a method for accurately detecting the objective quality of a video stream transmitted through a network. The hardware system used in the method includes: at the sending end, the video code stream objective quality evaluation device and the video code stream described in the embodiment of the present invention The stream sending device is connected, as shown in Figure 7; at the receiving end, the video code stream receiving device is connected with the video code stream objective quality evaluation device described in the embodiment of the present invention, as shown in Figure 7.

As shown in accompanying drawing 7 and accompanying drawing 4, the basic principle of the method described in the present embodiment is:

At the sending end, first, determine the video category and specific sequence used in the current test, and select the corresponding encoding control parameters according to the selected video compression method;

Then, use the video stream generation unit of the video code stream objective quality assessment device to select the corresponding original video sequence, and generate the video sequence number and image sequence number, superimpose on the original video sequence, and then use the corresponding video encoder to compress into a test-specific compressed video flow.

In order to save the time of video stream generation, various types of original video sequences can also be compressed into various types of video streams according to certain parameters in advance for storage. When in use, the corresponding video streams can be directly selected for output according to the video encoding control parameters. Finally, the video stream sending device will not The test-specific compressed video stream of less than 10s is cyclically sent to the network for transmission. The cycle sending time can be determined according to the specific test requirements, and the transmission format and protocol can be determined according to the specific network physical layer transmission protocol.

The method of the present invention can be adapted to the following situations: at the receiving end, first, the video stream receiving device is used to receive and restore the video stream sequence into a video stream sequence that can be processed by a video decoder according to a specific network transmission protocol;

Then, send it to the objective quality evaluation device of the video code stream for processing. Wherein, the video stream receiving and analyzing unit extracts the NALU unit of the video code stream and analyzes and extracts the basic parameters of the video stream, such as frame type, POC number, video resolution and quantization step size, etc. for subsequent analysis;

The video stream reconstruction and display unit decodes and stores the received compressed video stream and sends it to the display device for playback;

The video acquisition and frame information identification unit collects and stores the decoded video played on the display device, and extracts the information characters at the specified position to obtain the decoded video type and POC number for subsequent analysis;

The video objective quality calculation unit extracts the corresponding original video sequence and compressed video stream in the video stream generation unit according to the video stream parameters and various types of video sequences output by the aforementioned units, and calculates the objective quality evaluation result of the output video stream.

As shown in FIG. 5 , the method for extracting the corresponding original video sequence and compressed video stream includes: binary code stream comparison method and video frame information extraction method. The two types of methods can be used alone or together to authenticate each other, depending on the configuration of the detection environment.

Among them, the binary code stream comparison method needs to compress various types of original video sequences into various types of video streams for storage at the receiving end; or, after obtaining the received video stream type and encoding parameters, call the receiving end video stream generation unit to generate corresponding compression video stream. Specifically: the binary code stream comparison method utilizes the parameters extracted by the video stream receiving and analyzing unit to limit the analysis range of the compressed video stream at the receiving end. After obtaining a code stream structure of a test-specific compressed video stream of not less than 10s, the video objective quality calculation unit uses a binary comparator to compare each NALU unit of the code stream with the compressed video stream stored at the receiving end to obtain The compressed video stream that matches the received video code stream is used to extract the corresponding original video sequence and compressed video stream for the calculation of objective video quality. This method is more accurate.

The video frame information extraction algorithm directly extracts the video stream type and image sequence number from the decoded video sequence. Specifically: the video frame information extraction method extracts the video sequence number letter image superimposed on the specified position on the video frame, and performs correlation matching with the letter features in the character feature library to determine the letter information and the corresponding sequence number, and extract the corresponding original video accordingly. Sequence and compress video streams for use in video objective quality calculations. This method is susceptible to video degradation caused by network packet loss. If the reading is inaccurate, it is necessary to extend the observation time until the network returns to stability to obtain accurate recognition results.

Wherein, the step of calculating the objective video quality according to the received compressed video stream and the extracted corresponding original video sequence includes:

Decode and generate a video sequence from a received compressed video stream unit of not less than 10s;

Judging whether the POC number extracted by the aforementioned unit analyzes whether packet loss occurs in the compressed video stream;

If there is packet loss, it is necessary to make up the lost video frames in the decoded video sequence. The filling method can be directly used frame copy method or motion compensation method. The frame copy method directly copies the previous video frame to the current lost video frame position; the motion compensation method complements the information of the current lost frame position according to the motion compensation relationship of the previous and subsequent frames.

As shown in FIG. 6 , when the decoded video sequence is completed, the objective error between it and the corresponding original video sequence can be calculated. Full-reference video objective quality assessment parameters such as Mean Squared Error (MSE for short), Peak Signal to Noise Ratio (PSNR for short) and Mean Structural Similarity Index Method (MSSIM for short) can be used.

Assuming that the video frame resolution is M×N (pixels), X represents the original video sequence, and Y represents the decoded video sequence, the above evaluation parameters can be calculated as follows:

MSE can be obtained according to formula (10):

$\mathrm{<span\; class="notranslate">\; MSE</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; N\; M</span>}}\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; Y</span>}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; ]</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 10</span>}<span\; class="notranslate">\; )</span>$

Among them, X(i, j) is the pixel value of a certain frame of the original video sequence at point (i, j), and Y(i, j) represents the image of the corresponding frame of the decoded video sequence at point (i, j). pixel value.

PSNR can be obtained according to formula (11):

$\mathrm{<span\; class="notranslate">\; PSNR</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 10</span>}\mathrm{<span\; class="notranslate">\; log</span>}<span\; class="notranslate">\; [</span>\frac{\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; 255</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; Y</span>}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; ]</span>}<span\; class="notranslate">\; ]</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 11</span>}<span\; class="notranslate">\; )</span>$

Where X(i,j) is the pixel value of a certain frame of the original video sequence at point (i,j), and Y(i,j) represents the pixel value of the corresponding frame of the decoded video sequence at point (i,j). Pixel value, 255 is the peak amplitude of the video signal under 8bit sampling conditions.

MSSIM can be obtained according to formulas (12), (13), (14), (15), (16):

First, calculate the structural similarity index method (SSIM for short), the calculation expression is as follows:

SSIM＝[l(x,y)] ^α ·[c(x,y)] ^β ·[s(x,y)] ^γ (12)

Among them, α, β, γ>0, are the weighting coefficients of brightness comparison function l(x,y), contrast comparison function c(x,y) and structure information comparison function s(x,y) respectively. The calculation expressions of these three functions are as follows:

$\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 2</span>}{\mathrm{<span\; class="notranslate">\; \&mu;</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}{\mathrm{<span\; class="notranslate">\; \&mu;</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}}{{{\mathrm{<span\; class="notranslate">\; \&mu;</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{{\mathrm{<span\; class="notranslate">\; \&mu;</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; K</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 13</span>}<span\; class="notranslate">\; )</span>$

$\mathrm{<span\; class="notranslate">\; c</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 2</span>}{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}}{{{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; K</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 14</span>}<span\; class="notranslate">\; )</span>$

$\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; xy</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}}{{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; xy</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 15</span>}<span\; class="notranslate">\; )</span>$

in, represents the average brightness of the original video sequence, Indicates the average brightness of the decoded video sequence; Indicates the standard deviation of the original video sequence, ${\mathrm{\&sigma;}}_{\mathrm{the\; y}}={\left(\frac{1}{N-1}\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}{(x_i-{\mathrm{\&mu;}}_{\mathrm{the\; y}})}^2\right)}^{1/2}$ Indicates the standard deviation of the decoded video sequence; ${\mathrm{\&sigma;}}_{\mathrm{xy}}=\frac{1}{N-1}\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}(x_i-{\mathrm{\&mu;}}_{\mathrm{the\; y}})({\mathrm{the\; y}}_i-{\mathrm{\&mu;}}_{\mathrm{the\; y}})$ Indicates the covariance of the original video sequence and the decoded video sequence; C ₁ , C ₂ , and C ₃ are constants, and K ₁ , K ₂ <<1. L represents the dynamic variation range of the pixel value. When L=255, it means that the video image is an 8-bit image.

The quality evaluation value of the entire video frame is represented by a structural similarity mean (MSSIM, Mean Structural Similarity):

$\mathrm{<span\; class="notranslate">\; MS\; SIM</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; m</span>}}\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{\mathrm{<span\; class="notranslate">\; m</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; SSIM</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 16</span>}<span\; class="notranslate">\; )</span>$

Where M is the number of local windows (the size of the local window is 8X8) in one frame of video, and x _k and y _k are the contents of the kth local window video frame.

The step of calculating the video network transmission quality according to the received compressed video stream and the extracted corresponding locally stored compressed video stream is specifically: according to the POC number extracted by the aforementioned unit in a compressed video stream unit not less than 10s, analyze Whether there is any packet loss in the received compressed video stream. The packet loss rate in video network transmission can be calculated according to formula (17):

At the same time, the bit error rate of the video compression source in network transmission can be calculated according to formula (18):

Since the network status is not stable during the network transmission process, in order to ensure the accuracy of the test results, a large number of tests can be performed in a loop, and the above objective quality evaluation results of the video sequences generated by decoding a large number of compressed video stream units not less than 10s can be calculated. output after averaging.

Embodiment two:

This embodiment is an evaluation method for accurately detecting the compression quality of a video encoder in a video communication system. The hardware system used in the method includes: the video code stream objective quality assessment device described in the embodiment of the present invention is connected to a video encoder device, as shown in FIG. 8 .

As shown in accompanying drawing 8 and accompanying drawing 3, the method described in embodiment comprises:

First, the original video sequence is generated by the video stream generation unit in the video code stream objective quality assessment device according to the input test video category and parameters;

Then, send the sequence to the video encoding device to be evaluated to generate a compressed video stream;

Then, the compressed video stream is sent back to the video code stream objective quality assessment device, and after being processed by the video stream receiving and analyzing unit and the video stream reconstruction and display unit, a decoded video sequence is generated.

Finally, the video objective quality calculation unit refers to the original video sequence and the decoded video sequence to calculate full-reference video objective quality assessment parameters such as MSE, PSNR and MSSIM.

The above method can more conveniently realize the compression quality evaluation of the video encoder in the video stream generation stage in the video communication system.

Embodiment three:

This embodiment is an evaluation method for accurately detecting the video reconstruction quality of a video playback and display device in a video stream reconstruction stage in a video communication system. The hardware system used in the method includes: the video code stream objective quality assessment device described in the embodiment of the present invention is connected to an independent playback and display device, as shown in FIG. 9 .

As shown in Figure 3 of accompanying drawing 9, the method described in this embodiment includes:

First, the original video sequence is generated by the video stream generation unit in the video code stream objective quality assessment device according to the input test video category and parameters;

Then, the sequence is sent to the video stream reconstruction and display unit, and the unit controls an external independent video playback and display device to play the original video sequence;

Then, use the video acquisition and frame information identification unit to acquire, identify and store the video sequence obtained from an external independent video playback and display device;

Finally, the video objective quality calculation unit calculates full reference video objective quality assessment parameters such as MSE, PSNR and MSSIM with reference to the original video sequence and the stored playback video sequence.

The video number and image sequence number are superimposed in the process of generating the original video sequence, and the objective quality assessment of the video can realize the identification and alignment of the collected video sequence and the original video sequence, so that the video reconstruction stage in the video communication system can be realized more conveniently. Generated video quality assessment.

The present invention also provides a method for evaluating the quality of video streams, which can be applied to the sending end, including:

Obtaining an original video sequence, a sequence identifier corresponding to the original video sequence, and an image sequence number corresponding to the original video sequence;

An original compressed video stream corresponding to the original video sequence is generated according to the sequence identifier and the image sequence number.

The step of generating the original compressed video stream corresponding to the original video sequence according to the sequence identifier and the image sequence number is specifically:

The sequence identifier and the image sequence number are superimposed on the frame of the original video sequence to generate an original compressed video stream.

The present invention also provides a quality evaluation device for video streams, which can be set at the sending end, including:

An acquisition unit, configured to acquire an original video sequence, a sequence identifier corresponding to the original video sequence, and an image sequence number corresponding to the original video sequence;

The generating unit generates an original compressed video stream corresponding to the original video sequence according to the sequence identifier and the image sequence number.

The above description is a preferred embodiment of the present invention, it should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, some improvements and modifications can also be made, and these improvements and modifications can also be made. It should be regarded as the protection scope of the present invention.

Claims (10)

1. A method for evaluating the quality of a video stream, comprising: Obtaining a first compressed video stream generated by processing the original video sequence; the first compressed video stream carries a sequence identifier and an image sequence number corresponding to the original video sequence; Acquiring the original video sequence corresponding to the first compressed video stream according to the sequence identifier; Acquiring an original compressed video stream corresponding to the original video sequence according to the sequence identifier and the image sequence number; Evaluate the video quality of the first compressed video stream according to the first compressed video stream and the original compressed video stream. 2. The method according to claim 1, further comprising: Obtain a first video sequence corresponding to the first compressed video stream; Evaluate the video quality of the first video sequence according to the first video sequence and the original video sequence. 3. The method according to claim 2, wherein, before the step of evaluating the video quality of the first video sequence according to the first video sequence and the original video sequence, the method Also includes: Judging whether there is a frame loss in the first video sequence: if there is a frame loss, the missing video frame is completed to generate the first video sequence after the frame complement; The specific step of evaluating the video quality of the first video sequence according to the first video sequence and the original video sequence is: according to the first video sequence and the original video sequence after supplementary frames, Evaluate the video quality of the first video sequence after frame supplementation. 4. The method according to claim 1, wherein the processing of the original video sequence comprises: Coding processing of the original video sequence; transmission processing of a raw compressed video stream generated from said raw video sequence; or Decoding the received original compressed video stream generated according to the original video sequence. 5. The method according to claim 1, wherein the step of obtaining the first compressed video stream generated by processing the original video sequence is: Receiving the video of the original compressed video stream after transmission processing as the first compressed video stream; Receiving the decoded video of the original compressed video stream as the first compressed video stream; or A video displayed on the original compressed video stream is collected as a first compressed video stream. 6. The method according to claim 2, wherein the step of obtaining the first video sequence corresponding to the first compressed video stream is specifically: Decoding the first compressed video stream to generate a first video sequence; or Video displayed on the original compressed video stream is collected to generate a first video sequence. 7. The method according to claim 1, wherein the step of obtaining the original video sequence corresponding to the first compressed video stream according to the sequence identifier comprises: Extracting the sequence identifier carried by the first compressed video stream; According to the corresponding relationship between the sequence identifier and the original video sequence, the original video sequence corresponding to the sequence identifier is acquired. 8. The method according to claim 1, wherein the step of obtaining the original video sequence corresponding to the first compressed video stream according to the sequence identifier comprises: Extracting the sequence identifier carried by the first compressed video stream; According to the sequence identifier, an original video sequence corresponding to the sequence identifier is generated. 9. The method according to claim 1, wherein the step of obtaining the original compressed video stream corresponding to the first compressed video stream according to the sequence identifier and image serial number comprises: Extracting the sequence identifier and image sequence number carried by the first compressed video stream; The original compressed video stream corresponding to the sequence identifier and the image sequence number is acquired according to the correspondence between the sequence identifier, the image sequence number and the original compressed video stream. 10. The method according to claim 1, wherein the step of obtaining the original compressed video stream corresponding to the first compressed video stream: Extracting the sequence identifier and image sequence number carried by the first compressed video stream; The sequence identifier and the image sequence number are superimposed on the frame of the original video sequence to generate an original compressed video stream.